2024 Primary Election Study

Study Overview:
The 2024 Primary Election Study (PES) was organized by Sarah E. Anderson, Daniel M. Butler and Laurel Harbridge-Yong to study primary voters in the 2024 Senate primaries. The study surveyed representative state samples in CA, MI, and NV (1,000 respondents in each) and oversamples of likely primary voters in potentially competitive primaries (Democratic and Republican primary voters in CA and MI, Republican primary voters in NV). Likely primary voters (likely_voter_rev==1) are defined as respondents who indicated they are registered to vote and reported a value of 10 (out of 10) on their chance of voting in primary.

All respondents answered the common content questions and then were randomized into one of 5 team modules. This data release includes the common content. The teams' data will be released in fall 2026.

Vote choice reflects all candidates on the ballot (as of a few weeks before fielding began). Several questions (i.e., ideology, electability) ask about a subset of candidates. The PES focused on the leading candidates (those polling above 5% in recent polls and/or with evidence of viability in fundraising reports where polling was infrequent or inconsistent).

Files:

2024-001_methods-revised-9_30_25.pdf 	Methods report from survey provider (Verasight) with details on sampling, weighting, voter validation, and data quality.

PES 2024 codebook common content.xlsx	Codebook of variables (questions and response options) in the PES common content

PES 2024 Common Content.dta		PES data for common content question in Stata (.dta) format

PES 2024 Common Content.csv		PES data for common content question in comma separated (.csv) format

Edits to survey between CA and other states:
Added "request response" and reminder that respondents "must click on the slider to record your response" for the feeling thermometer and electability questions. In CA, because the feeling thermometer starting point was 0, a large fraction of respondents recorded values of "missing" for the out-party when they may have meant to select 0.
